August 24-25, 2006, Cologne, Germany
| International HL7 Interoperability Conference IHIC 2006
August 24-25, 2006, Cologne, Germany |
|
||
To the IHIC 2006 conference program committee
Email: shabo@il.ibm.com
Primary Presenter:
Name: Amnon Shabo (Shvo)
Affiliation: IBM Research Lab In Haifa
Address: Haifa University Campus, Haifa 31905, Israel
Phone: +972-544-714070
Fax: +972-4-8296116
Email: shabo@il.ibm.com
Co-Authors:
John Sharko, University of Massachusetts at Lowell
Brain Drohan, University of Massachusetts at Lowell
Kevin Hughes, Massachusetts General Hospital
Presenter bio:
Amnon Shabo (Shvo), PhD, works at IBM Research Lab in Haifa as a research staff member and is involved in various Healthcare & Life Sciences projects. He specializes in health information standards and is active in HL7 in the following roles:
- A co-chair of the Clinical Genomics Special Interest Group, as well as its modeling facilitator and primary contributor;
- A co-editor of CDA Release 2;
- A co-editor of the CCD (Continuity of Care Document) - a joint effort of HL7 and ASTM in applying ASTM CCR to CDA.
In addition, Amnon specializes in longitudinal and cross-institutional Electronic Health Records (EHR) and was a co-author of the mEHR proposal made by a consortium of nineteen partners to the European Commission's Sixth Framework Program, based on his vision of Independent Health Records Banks for addressing the lifetime EHR sustainability challenge.
Family History Information Exchange Using the HL7 V3 Clinical Genomics DSTU
research
30
A number of family history applications are in use by healthcare professionals and patients. Each has its own proprietary data format for family history health information, thus making it difficult to exchange information between programs. By using the HL7 V3 Clinical Genomics Family History DSTU, disparate family history applications will be able to exchange an individual’s family history. This standard holds the pedigree structure as well as an appropriate representation of clinical-genomic correlated data. We will describe an implementation project conducted by Massachusetts General Hospital, University of Massachusetts at Lowell and IBM Research Lab in Haifa.
Family History Information Exchange Using the HL7 V3 Clinical Genomics DSTU
Amnon Shabo (Shvo),John Sharko, Brain Drohan, Kevin Hughes
We will describe an ongoing implementation project where we have developed transformations between proprietary data formats of common applications, and potentially, genetic laboratories that screen for mutations in oncogenes and require family history data for genetic testing order. The transformations are based on the HL7 V3 Clinical Genomics Family History model. One of the use cases for clinical-genomics family history is in hereditary breast and ovarian cancer. These cancers are associated with variations in the BRCA genes. A complete family history can be assessed for potential risk for the patient (proband), supporting clinical decision making in genetic counseling and at the point of care.
Ethical and Legal Issues:
Family history data might involve accessing other family members' medical history, which raises ethical issues relating to privacy within families. In addition, patients are sometimes expected, or even encouraged, to enter their family history directly (.e.g., the US Surgeon General tool, see http://www.hhs.gov/familyhistory/), and the challenge arises as to how to incorporate patient-entered data into legally-authenticated medical records held by healthcare providers.
Data Representations:
While a pedigree is hierarchical by nature, sometimes the patient knows about a relative's disease but cannot recall the precise position of that relative on the pedigree (or only knows parental affiliation, i.e., relative is on the paternal or maternal side), and also there are possible intermarriages. Thus, even though the basic structure is hierarchical there is a need for flat representation as well. The possible representations we have considered are (1) pure hierarchical representation: the patient and all his/her relatives are only associated with their first-degree relatives; (2) pure flat representation: all relatives are directly associated with the proband and (3) mixed representation: some relatives are directly associated with the proband (mainly those whose position cannot be determined). Each of these alternatives has pros and cons but it seems that the third one is a flexible structure that enables the representation of all use cases. Its main downside is that it is a complex model with optionality that might create confusion in actual implementations.
HL7 Modeling considerations:
Using the pure flat representation requires the use of three ids for each person: self id and the person's mother and father ids. These ids are needed in order to reconstruct the pedigree graph, which is necessary for the risk assessment processes. Since each relative is represented by a Role class where the id attribute can have the cardinality of 0..3, it is considered non-conformant to capture the semantics of each id using the II (instance identifier) data type. Using pure hierarchical representation doesn't require the use of ids at all but the fuzzy data typically given by the patient cannot always constitute clean representations. The mixed approach is to use the hierarchical representation in order to represent the father and mother of each relative while using their Person id attribute (the Person who is playing the Role) while assuring their global uniqueness. Another modeling consideration relates to the family relations codes which can not be qualified by parental affiliation due to the CE data type of Role.code attribute.